When I use several actions in a row, it gives different answers

In the story part, cannot make the right decision after 4-5 actions and either does not give the rest of the answers at all or answers with other utters that I do not use. I’ve had this problem before, so is there a limit we should keep our stories? Does get confused when creates a story that is too long?


@grigrigrigon Hi, I must say this is not a right approach for your use-case (Hope you understand), can I request to create different stories for the different utterance, but yes you can take min 2 or 3 which are related else you should create the different stories for the same.

Read about the TED Policy, especially its max_history attribute:

This parameter controls how much dialogue history the model looks at to decide which action to take next. Default max_history for this policy is None, which means that the complete dialogue history since session restart is taken into account.

If you want to limit the model to only see a certain number of previous dialogue turns, you can set max_history to a finite value.

Please note that you should pick max_history carefully, so that the model has enough previous dialogue turns to create a correct prediction.

Here is how the config would look like:

- name: TEDPolicy
  max_history: 8

Furthermore, why are you doing this? There is probably a better way to do that, like merging utterances, or, if you can’t, using a single Custom Action that displays them:

def run():
    dispatcher.utter_message(response = "utter_foo")
    dispatcher.utter_message(response = "utter_bar")
    dispatcher.utter_message(response = "utter_baz")
    # etc...

    return []